Functional Roles and Career Paths in Wikipedia

Page created by Marjorie Stevens
 
CONTINUE READING
Functional Roles and Career Paths in Wikipedia
Functional Roles and Career Paths in Wikipedia
     Ofer Arazy                       Felipe Ortega                      Oded Nov                 Lisa Yeo                 Adam Balila
 University of Alberta             University Rey Juan                   New York             Loyola University          University of Haifa
  University of Haifa                      Carlos                        University               Maryland            adambalila@hotmail.com
ofer.arazy@gmail.com               felipe.ortega@urjc.es               onov@nyu.edu           mlyeo@loyola.edu

ABSTRACT                                                                            become involved in the project’s administration. Although
An understanding of participation dynamics within online                            on the surface peer production projects may be thought to
production communities requires an examination of the                               be non-hierarchical, in reality a core group of leaders
roles assumed by participants. Recent studies have                                  usually emerges (often through formal election processes)
established that the organizational structure of such                               to provide centralized coordination, manage quality control
communities is not flat; rather, participants can take on a                         processes, mediate conflicts, and develop organizational
variety of well-defined functional roles. What is the nature                        policies [4, 16, 43].
of functional roles? How have they evolved? And how do
participants assume these functions? Prior studies focused                          Online production communities are characterized by a core-
primarily on participants’ activities, rather than functional                       periphery structure, where the majority of contributors are
roles. Further, extant conceptualizations of role transitions                       involved in few tasks and participate at the community’s
in production communities, such as the Reader to Leader                             periphery, while a relatively small portion of contributors
framework, emphasize a single dimension: organizational                             take on additional responsibilities and constitute the core
power, overlooking distinctions between functions. In                               [12, 14, 20, 24, 30]. The core-periphery conceptualization
contrast, in this paper we empirically study the nature and                         concentrates on participants’ power (or authority) within
structure of functional roles within Wikipedia, seeking to                          the community, emphasizing lateral movements from the
validate existing theoretical frameworks. The analysis sheds                        community’s fringes to positions of responsibility and
new light on the nature of functional roles, revealing the                          influence [56]. Prior works in the area have tried to model
intricate “career paths” resulting from participants’ role                          the process of transitioning from the community’s
transitions.                                                                        periphery to the core. For example, the ‘Reader to Leader’
                                                                                    (R2L) framework [48] attempted to synthesize these earlier
Author Keywords                                                                     works and provide a comprehensive conceptualization on
Peer-production; Wikipedia; organizational structure;                               the successive steps volunteers take on their way to
functional roles; role transitions.                                                 community leadership. According to this framework, while
INTRODUCTION                                                                        the main path from the periphery to the core leads through
Recent years have seen the emergence of a community-                                these successive stages, alternative routes are possible,
based model for the production of knowledge-based goods                             including those representing attrition (i.e. transitions from
such as Wikipedia and open-source software [44]. As                                 the core to the periphery). Notwithstanding the importance
participants become more involved in their projects and                             of organizational power in characterizing the composition
gain the community’s trust, they gradually move from the                            of online communities, this conceptualization overlooks
periphery to the community core, gaining access to more                             other important dimensions, such as the functional
sensitive and influential decisions [1].                                            organization of community work. Thus, extant
                                                                                    conceptualization lack the capacity to distinguish between
The problem of motivation for participation in peer                                 different functions situated at the same level of the
production communities has attracted significant attention                          organizational structure. Traditional organizations are
at the academic community, and numerous studies have                                characterized by a functional organization, dividing work
discussed the various motives driving volunteers to                                 between departments such as: Finance, Operations,
contribute their time, effort, and expertise. Less is known,                        Marketing, etc., and the functional organization is at the
however, about how these volunteers sustain and increase                            core of scholarly literature on management and
their participation, take on additional responsibilities, and                       organizations [40]. However, the study of functional roles
                                                                                    has been largely absent from the literature on online
Permission to make digital or hard copies of all or part of this work for           communities.
personal or classroom use is granted without fee provided that copies are not
made or distributed for profit or commercial advantage and that copies bear this
notice and the full citation on the first page. Copyrights for components of this
                                                                                    Therefore, the objective of this study is to investigate the
work owned by others than ACM must be honored. Abstracting with credit is           functional organization of online production communities:
permitted. To copy otherwise, or republish, to post on servers or to redistribute   the nature of functional roles and the way in which these
to lists, requires prior specific permission and/or a fee. Request permissions
from permissions@acm.org.
                                                                                    roles are traversed when participants move from the
CSCW 2015, March 14–18, 2015, Vancouver, BC, Canada.                                periphery to leadership positions. In part, our study could
Copyright 2015 © ACM 978-1-4503-2922-4/15/03...$15.00.
http://dx.doi.org/10.1145/2675133.2675257
Functional Roles and Career Paths in Wikipedia
be viewed as an empirical investigation of existing                social (or organizational) settings and differ in terms of
conceptualizations of emergent leadership, namely the              their accessibility (the extent to which it is easy to accept a
Reader-to-Leader (R2L) framework [48]. Despite the                 role) and situational contingency (the contextual factors that
significant attention that the R2L framework has received          affect action). Roles are resources that help people
within the research community, to date, this framework has         accomplish their goals, and tools used in the establishment
not been empirically evaluated. In fact, Preece and                of social structure [8]. Hence, social (or organizational)
Shneiderman [48] acknowledge that “Even though our                 structure can be viewed as an ecology of roles.
[R2L] framework and discussion of its components are
                                                                   Formal organizations are generally understood to be
supported by research, the framework needs empirical
                                                                   systems of coordinated and controlled activities that arise
testing, which is an obvious next stage of this work” (p. 24).
                                                                   when work is embedded in complex networks of technical
The setting for our study is Wikipedia, one of the most            relations and boundary-spanning exchanges [38].
notable examples of peer production [9]. Wikipedia was             Organizational structure affects organizational action in two
able to recruit thousands of volunteers to produce millions        primary ways. First, it provides the foundation on which
of encyclopedic entries in 287 languages, and develop              standard operating procedures and routines rest. Second, it
extensive policies and mechanisms for governing its                determines individuals’ role assignment and their access to
collaborative authoring process. Wikipedia’s success               decision-making processes. Hence, the structure determines
attracted the attention of scholars, who investigated              the extent to which individuals at different organizational
Wikipedia’s organizational model [4, 24, 49]. Our research         positions can influence the organization’s actions [39].
methodology employs publicly available Wikipedia system
                                                                   ORGANIZATIONAL ROLES IN ONLINE COMMUNITIES
logs to identify participants’ access privileges. It builds on     Online communities, and in particular production
the framework from [3] to map access privileges onto               communities, have been investigated extensively in recent
organization roles, and analyzes role transition for a large       years. Relevant to our inquiry of roles and organizational
sample of Wikipedia contributors. In addition, we perform a        structure are prior works that studied community members’
qualitative analysis of a selected subset of editors, adding       roles. Previous investigations have discovered that users
depth and context to the quantitative data analysis.               often follow very distinctive patterns of activity, playing
Hence, the goal of this study is to address the following          roles in their online community [58]. Production
research questions:                                                communities are often described in terms of a core-
                                                                   periphery structure, which entails a dense, cohesive core,
RQ1: How are the functional roles of Wikipedia                     and a sparse, unconnected periphery [11, 13, 20, 33, 35, 45,
characterized and what activity profile is associated with         52]. Contributors play different roles: the majority of
each role?                                                         participants, who are not very active, are situated at the
RQ2: How has Wikipedia’s functional composition evolved            community’s periphery; and a small minority, who take on
over time?                                                         additional responsibilities and privileges, constitute the
                                                                   community’s core. For example, Long and Siau [35] found
RQ3: How do contributors to Wikipedia transition between           that social interaction patterns in open source projects start
roles on the path from community’s periphery to the core?          with a single hub, and as the project matures the
Our investigation yields important insights about the routes       configuration evolves into a core/periphery structure. Most
to leadership in peer production projects, informs extant          prior studies in the area have defined participants’ position
theory in this area, and offers guidelines to designers and        on the periphery-core continuum based on the quantity and
custodians of online communities.                                  types of activities they perform [5, 30, 34]. For example, [4]
                                                                   categorized contributors into two primary classes:
SOCIAL ROLES AND ORGANIZATIONAL STRUCTURE                          administrative- vs. content-oriented; and [57] applied a
The concept of social role has been the subject of extensive       combination of interpretive and network analysis methods
analysis in sociology [10]. Social roles encapsulate the           to identify key roles. Much less work has been devoted to
social context, history of actions, structures of interaction,     studying contributors’ formal duties [16]. A few prior
and the attributes people bring to the interaction by              works provide a partial description of Wikipedia’s formal
providing a meaning system, which both constrains and              roles [24, 42, 55]. Building on these earlier works, a recent
enables action [37]. The importance of this concept lies in        study [3] has analyzed access privileges in Wikipedia and
its utility: the classification of types of social relations and   developed a comprehensive description of Wikipedia’s
behaviors into a smaller set of roles reduces the analytic         organizational structure, including twelve roles (organized
complexity of social systems and facilitates the comparative       in a power hierarchy: Unregistered Users, Registered
study of populations across time and settings [32]. Roles          Users,     Manually       Registered      Users,    Technical
can be understood through two primary dimensions:                  Administration, Border Patrol, Quality Assurance, QA
structure and culture. The structural definition of roles          Technicians, Administrators, Security Force, Directors,
pertains to commonalities in behavior patterns, while the          Privacy Commissioner, and Benevolent Dictator (see Table
cultural dimension refers to roles that are recognized in          1).
Level on        Role                             Description                                            Access Privileges
Org. Chart
Level 0         Unregistered Users               Non-community members                                  *
                                                 Newly registered users                                 user
                Registered Users
                                                                                                        autoconfirmed
Level 1
                                                 New users who had to be manually registered to         confirmed
                Manually Registered Users        bypass some restrictions                               IPpblock-exempt
                                                 Privileged users responsible for the administration    filemover
                Technical Administration         of the technical aspects (e.g. user accounts, files)   accountcreator
                                                 Users responsible for fighting vandalism by            rollback
                Border Patrol
                                                 reverting malicious edits
Level 2
                                                 Privileged users responsible for patrolling            reviewer
                Quality Assurance                Wikipedia and ensuring content quality                 autoreviewer
                                                 Users who develop automated tools (i.e. edit           abusefilter
                QA Technicians
                                                 filters) to assist quality assurance work
                                                 Highly involved users that are responsible for the     sysop
Level 3         Administrators                   social administration of the English Wikipedia         bureaucrat
                                                 community
                                                 Highly trusted users who are working to keep           oversight
Level 4         Security Force                   malicious users out and combat intentional             checkuser
                                                 manipulations of content
                                                 Key users responsible for oversight of the             steward
                Directors                        Wikimedia organization                                 importer & transwiki
Level 5
                Privacy Commissioner             High-ranking users who investigate complaints          ombudsman
                                                 about violations of privacy policy
                Benevolent Dictator              Jimmy Wales; responsible for defining high-level       founder
Level 6                                          policies and norms and for overall direction of the
                                                 community

                              Table 1: The organizational structure of Wikipedia (adapted from [3])
ROLE TRANSITIONS IN ONLINE COMMUNITIES                             ways in which users participate in online communities and
Traditionally, when individuals advance their careers in           the various activities they take part in, no study offers an
organizations, they move through positions that provide            overall framework of career paths in online production
them with increasing degrees of responsibility, often              communities. Prior research has concentrated on
including vertical authority over others [14]. Recently,           descriptions of the core members’ (leaders, owners) duties
organizational theorists have argued that traditional notions      and activities [15, 46], the distribution of authority among
of career progression make several assumptions that no             these members [20], as well as on the factors affecting a
longer reflect workplace realities, such as: mobility within a     participant’s promotion to a leadership role [13, 17, 18, 33].
single large organization, stability in the organization and       For example, [46] showed that the nature of work differs
its environment, and hierarchical (i.e. vertical) progression      between Wikipedians and non-Wikipedians. Others have
[7]. Thus, there is a need to pay more attention to lateral (or    examined promotion procedures: [13] investigated the
horizontal) progression that relies on expertise, identity,        criteria for promotion decisions (i.e. the RfA process and
achievement, and community involvement, rather than on             adminship decisions); [19] studied the relationship between
hierarchy and formal authority [27]. Such lateral                  editors position within the social network and promotion
movements often involve transitions between functional             decisions; and [33] demonstrated that promotion depends
areas.                                                             on the candidate’s relative activity levels. These studies –
Online production communities are characterized by                 although rich in detail – provide a restricted and simplified
movements from the community’s periphery to positions of           conceptualization of the transitions towards leadership.
responsibility and influence at the community’s core [21,          Preece and Shneiderman [48] provide an overall framework
43, 56]. Although many studies have discussed the different        of emergent leadership in online communities by
synthesizing prior works in the area. They present a              the collaborative editing process, including a well-defined
nuanced classification of activities, organizing roles in         scheme of roles access privileges [14]. We focus on human
successive levels of involvement, termed the “Reader to           editors and exclude software bots [25] from this analysis. We
Leader” (R2L) framework. The reader category represents           note that in Wikipedia, anyone can contribute without
the lowest level of participation, and includes tasks such as     registering. This results in two types of participants:
browsing, surfing, and searching. Reading of user-                anonymous (represented by the ‘*’ tag), whose contribution
generated content posted by other participants can be             is recorded and associated with an IP address, and registered
thought of as legitimate peripheral participation [31]; it is a   users, whose contributions are associated with a user name.
necessary condition for knowledge reuse, and is a typical         Since we cannot track subsequent role transitions for
first step toward more active participation [47]. Next, the       participants who initially contribute anonymously (it is not
contributor category represents a gradual move towards            possible to univocally associate IP-based contributions with a
larger and more frequent contributions as participants’           user name when one starts as an anonymous contributor and
confidence grows and they feel empowered and appreciated          later registers an account), anonymous contributors were
[59]. Activities in this category include: contributing           excluded from our analysis.
comments and responses to others’ postings, rating previous
                                                                  Our mapping of access privileges onto organizational roles,
contributions, tagging and categorizing existing content,
                                                                  illustrated in Table 1, was based on the earlier work of [3],
and making small knowledge contributions (e.g. correcting
                                                                  employing a combination of top-down (review of relevant
an error, adding a hyperlink). The next level of
                                                                  scholarly literature and Wikipedia’s own definitions of roles
participation, collaborator, involves two or more
                                                                  and access privileges) and bottom up methods (using log data
contributors discussing, cooperating, and working together
                                                                  of all Wikimedia users and statistical analysis techniques). A
to create something or share information [22]. These
                                                                  key advantage of using the framework at [3] is that it
activities allow participants to bring more from themselves
                                                                  arranges Wikipedia roles on a continuum from the
and more fully express their opinions and ideas. Finally, the
                                                                  community periphery to the core, thus allowing us to analyze
leader category represents the highest level of participation.
                                                                  movements up (or down) the organizational chart.
In addition to being active contributors and collaborators,
                                                                  Transitions between roles were defined as follows: when a
leaders (also referred to as “owners”), are responsible for
                                                                  participant is assigned an access privilege that is associated
social management tasks (e.g. establishing community
                                                                  with a more advanced role, we designate her as being
norms and explicit policies or conflict resolution) and
                                                                  promoted to this role; and when a participant loses all
administration of the technology infrastructure [16]. A
                                                                  privileges associated with the advanced role, we designate
unique feature of the R2L framework is that it provides an
                                                                  her as being demoted to a less advanced role.
intricate picture of role transitions, acknowledging both
upstream and downstream transitions, as well as bypasses          Sample
of intermediate stages (e.g. from contributor directly to         Our sample included 2,174 Wikipedia editors who are all
leader, bypassing the collaborator stage). Notwithstanding        registered members. Since articles’ topical categories may
these merits, the R2L conceptualization seems to collapse         attract different types of contributors, we sought a sample of
two important dimensions: organizational power and                users who have contributed to a representative sample of
function, impeding the distinction between different              Wikipedia articles. Therefore, our sampling procedure used a
functional roles that are situated at a similar point on the      seed of 96 articles that provides a representation of
Reader–Leader continuum. We also note that, despite its           Wikipedia’s topical categories and which has been employed
influence in the CSCW community, the R2L framework has            in earlier studies [4, 6]. This set of articles (created in January
not yet been examined empirically. The objective of the           2007) was selected based on randomization and a stratified
current study is, thus, to try and fill in these gaps in the      sampling of Wikipedia’s topics, congruent with Wikipedia’s
literature, empirically evaluate role transitions and assess      top-level classification1 [29] (categories: culture, art, and
the extent to which the R2L conceptualization captures the        religion; math, science, and technology; geography and
reality of career paths within production communities.            places; people and self; society; and history and events).
                                                                  From this original set of 96 Wikipedia articles, 3 were
METHOD                                                            discontinued and their edit history is no longer available,
                                                                  leaving us with 93 articles.
Research Setting
The focus of this empirical investigation is the English          Our sample of Wikipedia participants included every editor
Wikipedia and its community of editors. We base our data          who contributed at least one edit to any of these Wikipedia
gathering and analysis on Wikipedia’s system logs;                articles (prior to the 2007 cut-off). For the reasons mentioned
harvesting these logs can reveal important insights about         earlier, we excluded anonymous contributors and software
members’ ongoing behavior in its natural setting.                 bots. Our procedure rendered a sample of 2,174 distinct

Over time, the Wikipedia community has developed a
comprehensive and detailed set of procedures for governing        1For a list of Wikipedia top-level categories, please refer to
                                                                  http://en.wikipedia.org/wiki/List_of_overviews
participants. We followed these participants from the time         In order to provide richer context and ground the overall
they registered with Wikipedia until June 2012, recording          analysis in additional data sources and individual user
every change in their Wikipedia roles during this period.          cases, we identified a list of seventeen editors that capture
                                                                   the various role transition patterns, focusing on unique
Data Collection
                                                                   cases (e.g. editors that transition particularly quickly or
In order to address RQ1, we employed data harvested from
                                                                   slowly; extremely active editors, cases of demotions).
Wikipedia logs. To profile the activities of the various roles,
                                                                   Altogether, we investigated 17 editors with IDs: 11; 96;
we retrieved data regarding the contributions of our user
                                                                   20134; 29678; 44020; 82835; 124324; 126457; 274040;
sample across all Wikipedia namespaces, including: main
                                                                   889851; 1403682; 1812441; 1862829; 2164608; 2267145;
article pages (production work); talk pages (coordination
                                                                   3162157; and 3138762.We gathered information about
work); user and user talk pages (communicating with others);
                                                                   these editors from a variety of Wikipedia sources, including
Wikipedia and other pages (community work) [17]. For the
                                                                   editors’ personal pages and talk pages; requests for
second RQ, we gathered data regarding the evolution of roles
                                                                   adminship (RFA)5; requests for de-adminship6; the logging
within Wikipedia from an internal report by a senior
                                                                   table (e.g. checking who is operating a bot)7; and pages
Wikipedian2. Finally, for tackling RQ3 and collect data
                                                                   tracking dispute resolution cases. This selected set of
regarding role transitions, we first queried the Wikipedia API
                                                                   editors captured cases to exemplify the various role-
(http://en.wikipedia.org/w/api.php) to determine whether a
                                                                   transition dimensions; in terms of seniority: 8 veterans
participant is a registered user; we then used the API to
                                                                   (joining Wikipedia up to 2004) and 9 that joined later;
receive the list of role changes for each participant in our
                                                                   editing activity: 7 with relatively low activity patterns
sample, resulting in 7,563 role-change events. Organizing the
                                                                   across the various namespaces (no more than few thousand
role change history, we determined the participant’s role at
                                                                   edits), 5 moderate (tens of thousands of edits), and 5 highly
each point in time, and recorded role transitions.
                                                                   active editors (more than one-hundred-thousand edits);
In order to verify the accuracy of the user-role data, we          rank: 1 editor that never made a role transition, 14 that have
compared it to other sources: the rights log history3 and          gained Level 2 privileges, 13 Level 3 cases, and 1 editor at
alternative sources archiving role transitions at the early days   Level 5; pace: 5 that made fast transitions (less than a year),
of Wikipedia4. In cases of mismatch (7% of the events), a          5 that transitioned at a regular pace (one-three years), and 6
discrepancy was flagged, and was resolved manually. Some           slow movers (transitioning after more than three years); and
of the more common discrepancies included software bots            controversial: 5 editors that have been blocked (some only
that were initially registered as regular users (and once          temporarily and others for prolonged periods) and 7 that
identified as bots, were excluded from the sample) and             have been demoted (either voluntarily or forced by the
administrators who lost their privileges due to inactivity (in     community). We used these cases to illustrate role
such cases, we tracked the missing date for this event). In        transitions and career paths from a micro perspective.
addition, in a few cases transitions that were recorded
properly reflected a technical issue, rather than a role change    RESULTS
(e.g. a particular privilege was added, removed a few minutes      Out of the 2,174 editors in our set the majority were from
later, and then a different privilege was assigned). After a       the community’s periphery (i.e. auto and manually-
careful manual inspection of short-duration event sequences,       registered users), relatively few were at the intermediate
we excluded those events that did not seem to represent an         levels, and no instances were recorded for the top levels
actual role transition. Once we compiled a reliable list of        (namely, the ‘steward’, ‘importer & transwiki’,
access privilege transitions, we mapped them onto Wikipedia        ‘ombudsman’, and ‘founder’ privileges)8. In total, we
roles (as specified in Table 1 above) to arrive at role            recorded 21 Manually Registered Users (1 confirmed and
transitions. Many of the access privilege transition events did    21 IPpblock-exempt); 12 Technical Administrators (9
not induce changes in Wikipedia roles (e.g. a ‘filemover’          filemovers and 5 accountcreators); 127 users with Border
taking on the additional privileges of an ‘accountcreator’;        Patrol (i.e. rollback) privileges; 287 at the Quality
both corresponding to the Technical Administrator role). We        Assurance role (244 reviewer and 150 autoreviewer); 49
focused our attention on those access privilege transitions        QA Technicians (i.e. abusefilter); 436 Administrators (435
that entail a move between role categories.

                                                                   5
                                                                     en.wikipedia.org/wiki/Wikipedia:Successful_requests_for_admin
                                                                   ship
                                                                   6
                                                                      en.wikipedia.org/wiki/Wikipedia:Requests_for_de-adminship
2http://en.wikipedia.org/wiki/User:NoSeptember/The_NoSeptemb       7 For members joining Wikipedia prior to September 7, 2005,

er_Admin_Project                                                   some information is missing from the logging table; in those cases
3https://meta.wikimedia.org/w/index.php?title=Special%3ALog&t
                                                                   we used the time of first edit as a proxy for registration date.
ype=rights                                                         8 Across the entire Wikipedia community (all languages), there are
4
  Namely the Meta:Bureaucrat log                                   only 46 participants with the ‘steward’, ‘importer & transwiki’,
(https://meta.wikimedia.org/wiki/Meta:Bureaucrat_log), which       ‘ombudsman’, or ‘founder’ access privileges. This explains why
records privilege changes prior to 10 December 2004.               they were almost absent from our sample.
sysops and 10 bureaucrats); 11 users as Security Force (9          meta pages which pertain to community activity, such as
oversight and 9 checkuser); and 1 Director (with importer          talk, user, user talk, Wikipedia or Wikipedia talk pages.
privilege).                                                        Level 2 roles, on the other hand, present very high levels of
                                                                   activity at very specific namespaces linked to their
To address RQ1 concerning the activity profile of each role,
                                                                   particular duties (e.g. editors in Technical Administration
we partitioned each editor’s trajectory into different
                                                                   role spend substantially higher percentage of their work on
temporal sections, each listing the roles held in that time
                                                                   category pages, and QA Technicians focus more on
period, as well as the length and activities across the
                                                                   template pages).
various Wikipedia namespaces. Then, for each role, we
aggregated the activity of all editors. In terms of overall        To control for the effect of topical categories, we
activity, Technical Administrators and QA Technicians              partitioned the sample into sub-samples (by building on the
have the highest average daily activity, with 113 and 42           topical organization of the articles that were used to seed
daily edits respectively. Level 2 and 3 roles make on              the editor sample), and analyzed the activity patterns of
average 11-14 edits; while those at the entry (Level 1) and        each topical sub-sample. A multivariate analysis of variance
top levels (Levels 4 and 5) make less than 5 daily edits. In       (MANOVA) revealed very similar activity patterns across
terms of effort across the various Wikipedia namespaces,           topics – for all Wikipedia namespaces – suggesting that the
although all roles focus primarily on editing main pages,          results are independent of particular topical domains.
each role is characterized by a distinct activity profile, as
                                                                   Our next analysis focused on the evolution of roles,
illustrated in Figure 1. For example, New Registered Users
                                                                   addressing RQ2. Figure 2 illustrates the temporal evolution
focus primarily on editing main pages (74% of their edits);
                                                                   of roles within Wikipedia. As the figure shows, in the early
Technical Administrators and to a lesser extent Border
                                                                   days of Wikipedia, the entry level positions (Levels 0 and
Patrol are relatively inactive on main pages, but instead
                                                                   1: unregistered and registered members) were joined by
concentrate on coordination work (40% and 21%
                                                                   ‘core’ (or leadership) roles: Level 3 (sysop; 2002); Level 4
respectively in talk pages); Security Force is characterized
                                                                   (checkuser; 2005); Level 5 (steward; granted to the founder
by little production work (35% on main pages), and greater
                                                                   Wales in 2002); and Level 6 (founder; 2002). In 2004
focus on pages associated with policy creation and
                                                                   steward privileges were granted to other beyond the
enforcement (37% and 22% respectively on user talk and
                                                                   founder, whereas in 2006 the privileges of oversight
Wikipedia namespaces); and Directors are highly active on
                                                                   (Level4), ombudsman, importer, and transwiki (Level 5)
Wikipedia and Wikipedia talk namespaces (22% and 19%
                                                                   were introduced. Only at the next phase of Wikipedia’s
respectively). For example, editor #3162157, who joined on
                                                                   evolution - in 2008 and 2009 (with later additions in 2011)
January 2007 and never transitioned beyond the Registered
                                                                   was the intermediate functional layer introduced and Level
User role, made 301 edits to main pages; editor #274040,
                                                                   2 roles emerged.
who joined on May 2005 and gained various Level 2
(Technical Administrations, Quality Assurance, Border
Patrol) and Level 3 privileges, made roughly 30,000 edits
to main pages, 3,700 in talk pages, 25,000 in user and user
talk pages, and 14,000 in other namespaces; and editor #96,
a Director that joined at 2001, made 4,200 edit to main
pages and 1,000 edits across other namespaces.

    100%
     80%
     60%
                                               Others
     40%
                                               Wikipedia & Talk
     20%
      0%                                       User & Talk

                                               Main Talk

                                               Main
                                                                     Figure 2. The Evolution of Wikipedia’s access privileges

                                                                   In order to understand how these various roles were
                                                                   populated as Wikipedia evolved, we studied the distribution
                                                                   of access privileges by plotting the levels reached by the
Figure 1. Activity profiles associated with functional roles (as   users in our sample. In the early years of Wikipedia, the
            proportion of roles overall activity).                 vast majority of users were al Level 1; by 2005, as the
                                                                   second phase of Wikipedia’s life began, the project
Overall, we notice that editors at Level 3 and above               increasingly gained traction, and the number of new editors
consistently present a significant number of contributions to      joining rapidly grew; realizing the need to curate the
immense amounts of new content being constantly created,              2), where there are transitions between the Border Patrol
Level 2 functions were introduced and gradually populated.            and Quality Assurance roles. For example, editor #20134
Figure 3 shows the changes in the proportion of editors (y-           registered with Wikipedia in August 2003, gained the
axis) in each of the top five levels. We can see that the             Quality Assurance role in June 2010, and six months later
proportion of Level 3 users (sysop, bureaucrat) started to            added the Border Patrol role. Vertical transitions from one
represent a noticeable fraction of the total population circa         organizational level to another include transitions such as
April 2005, and from January 2008 it has remained quite               the one from the Registered Member (Level 1) to Quality
stable. Furthermore, since the introduction of levels in the          Assurance (Level 2) roles, or from Border Patrol (Level 2)
functional layer (Level 2), starting in January 2008, the             to Administrators (Level 3) (for example, editor #1862829
proportion of users taking roles from this layer has rapidly          registered on July 2006, gained the Border Patrol role on
increased, reducing the proportion of entry level users               April 2008, and became an Administrator on October
(Level 1). This finding offers evidence of the emergence of           2008). Overall, the majority of transitions were vertical
a group of intermediate level users focusing on tasks such            (995 instances; 90%). For the horizontal transitions, it is
as content curation and fighting vandalism.                           interesting to note that while there are some roles with little
                                                                      to/from transition (e.g. QA Technicians), other roles are
                                                                      characterized by frequent horizontal movement (e.g. Border
                                                                      Patrol)10.
                                                                      For the vertical transitions, while in some cases the
                                                                      transition to the community’s core is sequential. (e.g., 346
                                                                      transitions from Level 1 to 2 and 21 transitions from Level
                                                                      2 to 3), the majority of vertical transitions skip intermediate
                                                                      levels (e.g., 421 transitions from Level 1 to 3). Examples
                                                                      from our selected set of editors could illustrate the upward
                                                                      transition process: editors #2267145, #11, and #126457
                                                                      who made the transition to administrative role (i.e. Level 3)
                                                                      prior to the introduction of the functional layer (at 2007,
                                                                      2006, and 2005 respectively) moved directly from being
                                                                      Registered Users to administrative position, the latter two
                                                                      editors adding on Level 2 roles at a later time; in contrast,
    Figure 3. The proportions of role population over time.           editors #1862829 and #44020 attained their administrative
                                                                      position later (at 2008 and 2010 respectively) and have
An example from our selected subset of editors could                  made linear transitions: Level 1 => Level 2 => Level 3.
illustrate this evolution. Consider editor #126457, a
                                                                      Despite the majority of vertical transitions are upward (792
librarian who worked at Cambridge College and the British
                                                                      cases), a relatively large number of transitions were
Antarctic Survey, and primarily contributes to pages
                                                                      downward (195 cases). In particular, we noted 81 cases
covering historical figures, events, and places (in addition
                                                                      where core community members (Level 3) shed their
to his community activity). Editor #126457 became a
                                                                      special privileges and become regular members. For
member on October 2004, an Administrator (Level 3) on
                                                                      downward transitions, we distinguish between cases where
November 2005, and added the role of QA Technician
                                                                      an editor was first granted the higher level role and later
(Level 2) on July 2009.
                                                                      added a lower level role (as in the case of editor #126457),
Once we established the nature and evolution of                       and cases of demotion where an editor loses the higher level
Wikipedia’s functional roles, we turned our attention to              privilege (for example, editor #29678, who gained
investigating editors’ role transitions. In order to address          Administrator privileges on March 2009, lost them due to
RQ3, we first counted the times each role transition type             inactivity on October 2011). Table 2 presents these various
occurred in our sample. In total, we recorded 1,103 role              role transitions.
transitions (see Table 2). Out of the 2,174 participants, 732
                                                                      In order to represent the intricate role transition dynamics,
(34%) transitioned to various organizational roles, while the
                                                                      Figure 4 illustrates promotion and demotion role transitions
remainder 1,442 (66%) never moved beyond the entry
                                                                      (for clarity, we excluded from the figure same-level
levels (i.e. registered users)9.
                                                                      transitions, as well as cases of downward transitions that do
We observed both horizontal and vertical transitions. We              not entail demotion).
note horizontal transitions at the intermediate level (Level

9Note that after becoming an Administrator, if one adds a series of
                                                                      10
Level 2 roles, each such addition is recorded as a transition from      Cases where an editor played multiple Level 2 roles and lost one
Level 3 to the Level 2 role.                                          are not captured in our analysis.
Level                                            1                                                         2                                          3                4                5
                                     To

                                                                             Manually Registered

                                                                                                                                        Quality Assurance
                                                      Registered Users

                                                                                                                                                                              Administrators

                                                                                                                                                                                               Security Force
                                                                                                   Tech. Admin.

                                                                                                                   Border Patrol
                Level

                                                                                                                                                              QA Tech.

                                                                                                                                                                                                                Directors
                                                                             Users
                                    From

                        Registered Users                                             10             +2            +109                 +217                   +16            +420               +4              +1
                  1
                        Manually Registered Users                 7                                                                          +2                                   +1
                        Technical Administration       -1
                        Border Patrol                 -14                                5                6                                   66                             +16
                  2                                   -14                                2                6            12                                                     +4
                        Quality Assurance
                        QA Technicians                       -2                          1                                                           1                            +1

                  3     Administrators                -81                                6                4       8 / -7               4 / -3               40 / -2                             +7
                  4     Security Force                                                                                                                                   1
                  5     Directors

Table 2. Role transitions count; a plus sign (“+”) represents promotions, and minus sign (“-”) demotions; no sign indicates a same-
              level transition, or cases where higher-level roles were retained while picking up lower-level privileges.

Figure 4. Promotion and demotion transitions. Circle size corresponds to the number of editors in the sample; upward solid arrows
represent transition to a higher level; downward dotted arrows represent cases where editors lost higher level privileges. Width of
 arrows represents number of role transitions count. For clarity, same-level transitions, as well as downward transition that do not
                    entail demotion (i.e. editors kept their higher-level privilege), are excluded from the figure.
The analyses described above highlight the importance of               namespaces is also essential to achieve a promotion.
the mid-level strata of contributors. These contributors are           Timing also seems to have influenced the promotion
characterized by functional focus and diversity and                    process, and our qualitative analysis reveals the very low
represented by Level 2 roles in Wikipedia’s                            volume of participation in older voting processes for
organizational structure. In order to fully understand how             Administrator (in contrast with the high community
these roles are attained, we performed an analysis of                  participation from 2007 onwards). An extreme example is
incoming transitions into Level 2 roles (see Figure 5).                editor #96, who was made an Administrator on Feb. 2004
                                                                       with just 6 positive votes.
      100%
       90%                                                             The in-depth examination of 17 editors also shed
       80%
       70%
                                            Level 3&up: demotion
                                                                       additional light on the process of downward transitions.
       60%
       50%                                                             As revealed through the statistical analysis, a downward
       40%
       30%
                                            Level 3&up: adding roles
                                                                       transition can entail a demotion (when higher-level
       20%                                  Level 2: QA Tech           privileges are shed; e.g. editor #29678) or a case where a
       10%
        0%                                  Level 2: QA                higher-level role - typically, Level 3 - adds lower-level
                                            Level 2: Border Patrol
                                                                       privileges - commonly Level 2 (e.g. editor #126457).
                                                                       When studying the reasons for demotions, we found
                                            Level 2: Tech Admin
                                                                       evidence for both community-initiated and self-
                                            Level 1                    demotions. Reasons for community-initiated demotions
                                                                       were prolonged inactivity period (as in the case of editor
                                                                       #29678) or abuse of Wikipedia norms (often abuse of
Figure 5. Sources of incoming transitions into Level 2 roles.          rollback privileges). For example, editors #889851,
                                                                       #274040 and #124324 lost administrative privileges due
This analysis reveals that in addition to the differences in           to misuse of administrative tools, failing to respond to
terms of activity profiles reported earlier, the various               community concerns, or inappropriate off-wiki behavior.
Level 2 roles seem to represent different career paths. For            Self-demotions are often the result of continuous
example, transitions into Technical Administration are                 controversies and confrontations (e.g. editors #82835,
primarily through other Level 2 roles (namely, Border                  #1403682, and #2164608). We note that demotion cases
Patrol and Quality Assurance); progression into Border                 are often linked to blocking incidents, where an editor is
Patrol is directly from Level 1; and the role of QA                    prevented from editing activities for some period (in few
Technicians is often appended to Administrators.                       cases, indefinitely). For instance, editor #274040, who
In addition to the patterns emerging from our quantitative             was forced to give up administrative privileges, was
analysis (as discussed above), the qualitative analysis of             banned 12 different times (due to page-move vandalism,
the selected 17 editors revealed some interesting insights.            blocks and violations of earlier bans). In fact, regaining
Although we cannot validate statistically the significance             administrative privileges required a clean record of
of these findings, they do highlight some interesting                  ‘blocks’. It is interesting to note that extremely active
patterns. The discussions associated with promotion                    users who contribute across the various namespaces, are
decisions reveal that editing activity is a key consideration          often involved in controversies and disputes, and in
(see [13, 33]). As we have shown earlier, while all editors            multiple occasions have been banned.
focus their editing activities on main pages (see Figure 1),           Our qualitative analysis also revealed that some of the
administrators consistently present high activity levels in            users at higher ranks make use of advanced tools for
that namespace. The administrators in our representative               automating editing tasks (that is, they operate ‘bots’),
set - #1862829, #44020, #2267145, #11, #889851,                        leading to unusual peaks in editing activity. Most notably,
#126457, #29678, #274040, #124324, #82835, #1403682,                   we can identify the use of the AutoWikiBrowser tool as
#2164608, and #1812441 - had an average of 60,000 edits                one of the main triggers for an unusually high number of
in main (ranging 5,000-120,000) up to our study’s cutoff.              edits. For example, editor #1862829 is not only one of the
In addition, contributions to other namespaces, which                  developers of AutoWikiBrowser, he is also one of the 20
represent communications between editors and                           most active Wikipedians and runs a bot to carry out tasks
community activity, also seem to be relevant for                       that would otherwise be tedious.
promotion. For example, editor #2267145 has made over
50,000 edits to talk pages; #889851 has edited user and                Our study of the selected set of editors also examined
user Talk pages over 36,000 times; and editor #82835                   participation in additional community activity. When
made over 82,000 edits to talk pages, 38,000 to user and               analyzing participation in WikiProjects – projects
user talk pages, and roughly 85,000 other meta pages.                  typically dedicated to improving articles in a particular
This data suggests that while a significant editing activity           domain – we were not able to find evidence linking
in main pages is required for promotion to administrative              WikiProject participation to any particular user profile.
position, active participation in a variety of other
DISCUSSION                                                     periphery to the core of the community (with some
Recent years have seen increasing interest in investigating    additional, non-linear secondary paths that may skip a
organizational roles within online communities and the         particular stage), findings from the present study suggest
process by which contributors transition between roles.        that these non-sequential paths are the rule rather than the
Conceptualizations such as the Reader-to-Leader                exception. Namely, in most cases participants move
framework [48] have received significant attention. Yet,       directly from the entry levels to the community’s core
to date, there has been no quantitative empirical              (skipping intermediate functional roles). This may be an
evaluation of role transition processes. Our study of          artefact of the late introduction of Level 2 roles, although
career paths in Wikipedia over a period of over ten years      even after 2009 most of the Wikipedians arriving at Level
provides insights into the dynamics of transitions between     3 did so directly, bypassing Level 2 roles. The implication
the community’s periphery and core. The results highlight      of this finding is that Level 2 roles must not be seen as a
the significance of functional roles for understanding how     step towards becoming a community leader; instead, they
online production communities organize their work. In          represent functional positions that are important in their
particular, our findings call into focus the functional        own right.
middle organizational level (i.e. Level 2) of Wikipedia.
                                                               While most of the prior studies in the area emphasize
The findings show that formal roles determine users’           upward transitions towards leadership positions, our study
actual activity patterns across a variety of editing tasks     provides evidence for the existence of transitions from the
(i.e. namespaces). To date, much of the research on the        core back to the community’s periphery. In some cases
organizational structure within production relied on           this may simply reflect community regeneration, where
activity counts for determining one’s position on the          old leadership makes way for new generations of
periphery-core continuum [5, 30, 34].                          contributors that step-up to take additional responsibilities
A second contribution this study makes is in delineating       (as has been observed in the context of open-source
the evolution of Wikipedia’s organizational structure.         software development [52] and Wikipedia [49]). In fact,
Earlier studies have noted Wikipedia’s increased               in mid-2011 Wikipedia developed a policy to retire
bureaucracy [14] and have associated the growing               inactive administrators11. In other cases, however, these
complexity of Wikipedia’s organizational structure with        ‘downward’ transitions are the result of heated
decreased effectiveness [26, 54]. Our findings, on the         confrontations between community members. For
other hand, suggest that the introduction of additional        example, consider the case of editor #124324, who
functions (namely, Level 2 roles) was necessary for            engaged in arguments with a few other editors, eventually
curating the mounting number of new contributions. This        became frustrated with his treatment (both directly and
finding is in line with recent works which suggest that        through the formal arbitration/incident handling process),
online production communities that fail to create such         and voluntarily gave up his administrator privileges. In
organizational structures are not sustainable [41].            the words of editor #124324:

We also make a contribution in describing role transition           “In response to the Committee's decision to declare
dynamics within Wikipedia. Our empirical evaluation                 finding … “[#124324] ... has used his administrative
validates prior knowledge in the area [13, 23, 35, 43] and          tools while involved …)”, I've requested a desysopping”
demonstrates that the number of community members                   “anybody who cares: my self-block wasn't some kind of
decreases as we grow closer to the core: only a relatively          “tantrum” or strategy; it was a genuine attempt to get
small number of contributors proceed to Level 3 and even            the hell away”
fewer continue to becoming a core member (i.e. Levels 4
and higher). Level 2, because of its late introduction, is     In other cases, editors are forced to give up privileges, as
still not fully developed, but we expect that the number of    in the case of the editor #82835, one of the most active
contributors at this level would grow and surpass the          Wikipedians. The case involves accusations of disruptive
number of administrators (Level 3). Our results (see Table     editing against the editor. The demotion decision centers
2) clearly show that not all paths are equally traversed; in   around principles of collegiality and the use of automation
fact, some paths are never traversed (at least in our          tools. The community decided to revoke administrator
sample), suggesting that – although not formally               privileges and ban editor #82835 for up to one year.
articulated – there are de-facto career paths within           Interestingly, as with the upward transitions, we found
Wikipedia.                                                     that when core members (at Level 3 of the organizational
A key finding of our study is that the strength of paths (in   chart) move down the organizational ladder they most
terms of the frequency in which participants traverse          often transition directly to become a regular member,
them) does not correspond to the predictions of extant         rather than traversing through the intermediate Level 2.
conceptualizations. In particular, against the predictions
of the R2L model, which suggests a primary path to
                                                               11
leadership characterized by a linear sequence from the              http://en.wikipedia.org/w/index.php?title=Wikipedia:INACTIVITY
To the best of our knowledge, demotions were not                 person-focused, aversive, and legitimate leadership), and
previously investigated in the context of production             showed that these leadership behaviors are not restricted to
communities and no prior studies have investigated the           those in formal administrator positions; and [61] developed
departure – whether voluntary or compulsory – of key             a machine learning technique to automatically identify
community members (aside from the well-known case of             these leadership behaviors. [36] have analyzed leadership
Wikipedia cofounder Sanger’s exit of Wikipedia [53]).            behaviors in an online community, argued that these
While such demotions are the exception, rather than the          behaviors should be distributed between many community
norm, exposing this pattern enriches our understanding of        members (as opposed to centralized leadership), and
career paths within online production communities.               developed a tool to assist in leadership tasks. Our study
                                                                 extends these lines of investigation in four important ways:
Another insight from our study is that the dynamics of the
                                                                 (a) our study focuses contributors’ functional roles (as
roles transitions are more complex and intricate than
                                                                 opposed to leadership behaviors), and brings into attention
originally perceived: we found evidence for both horizontal
                                                                 the intermediate “professional” level that is often absent in
and vertical transitions; both upward and downward
                                                                 studies of leadership; (b) we study Wikipedia’s entire
movements; sequential progression through intermediate
                                                                 functional organization (and transitions between functional
phases, as well as direct transitions that bypass intermediate
                                                                 roles), rather than focusing on a particular type of
levels. Preece and Shneiderman [48] have already
                                                                 transition, thus enabling us to validate conceptualizations
postulated that role transition is not one-directional, and
                                                                 such as the Reader-to-Leader framework, and revealing
that users can move up and down the leadership ladder.
                                                                 unpredicted patterns (e.g. editors moving directly to
However, the evidence presented here points to complex
                                                                 administrative position and later assuming functional
patterns of transitions, whereby participants switch between
                                                                 lower-level roles; (c) we demonstrate that despite the loose
organizational roles (a pattern brought to the extreme by
                                                                 governance of Wikipedia and the fact that editing activities
one editor whose career spanned six role transitions). For
                                                                 are open to all, the functional role seems to determine the
example, we observed repeated cases where those who
                                                                 type of activities editors engage in, such that each
directly progressed to higher-level administrative roles (i.e.
                                                                 functional role is associated with a distinct activity profile;
Level 3) choose to later add intermediate functional roles
                                                                 and (d) we shed new light on downward transitions out of
(Level 2); a pattern not discussed in prior works. Not only
                                                                 the community’s core, a topic that has been under-
do we expose previously unobserved transition patterns,
                                                                 investigated.
our findings also offer empirically-validated weights (or
probabilities) for the various role transitions.                 Finally, we make a secondary methodological contribution
                                                                 regarding the usage of system logs as a tool for answering
In sum, our study builds on and extends prior knowledge in
                                                                 organizational research questions. Our experience has
the area in several important directions. In recent years,
                                                                 taught us that beyond the standard data cleaning that is
researchers in the area have begun to recognize the need to
                                                                 necessary, there is a need to triangulate data from multiple
move beyond the simplistic conceptualization of core-
                                                                 sources to ensure its reliability. Some of the examples we
periphery. One line of research has focused on the process
                                                                 have encountered include: automated bots not marked as
of promotion within Wikipedia. These studies tend to focus
                                                                 such; incomplete registration date for early entrants;
on a particular promotion decision, often to 'sysop': a
                                                                 inconsistencies between two sources recording granting of
single transition in the organizational chart and a single
                                                                 access privileges; missing data on role granting in
event in a user's trajectory. For example, [13, 17, 18]
                                                                 privileges logs (as evident by comments on editors’
studied how activity patterns affect promotion to
                                                                 personal pages); and irrelevant privileges granting data in
administrator position. While such studies are useful in
                                                                 the logs that do not reflect role changes (i.e. a series of
shedding light on why and how people move to
                                                                 erroneous transactions and their corrections). We would
administrative positions, they tell us little about the
                                                                 like to offer a word of caution to researchers relying on log
organizational structure in its entirety; further, they do not
                                                                 data in studying online communities, and to stress the
capture contributors’ full role transition trajectories. In
                                                                 importance of data triangulation.
addition, most of these studies view career path on a single
dimension - promotion to position of power – not                 Implications for design and management
distinguishing between the functional roles played at each       The findings of this study have important practical
position. Because of the focus on promotion to higher-order      implications for designers and administrators of online
positions (i.e. Administrator), they miss on the complex         communities. These communities strive to sustain
functional organization at the middle level of the               contributors’ participation, engagement and commitment.
organizational chart (i.e. Level 2). Another interesting line    An understanding of the paths contributors take could help
of research has been investigating the nature of leaders (and    to develop diverse “career paths” within the community,
leadership) within online communities, moving beyond the         such that contributors with different skill sets and interests
single-dimension characterization of leadership (power or        could find suitable avenues for channeling their energy. An
authority; i.e. core-periphery continuum). For example,          implication would be to place more emphasis on early
[60] investigated editors leadership styles (transactional,      detection and encouragement of contributors who seem to
be suitable (as indicated by their editing profiles) for         within Wikipedia based on editors access privileges. This
particular roles. Community owners could provide more            approach has advantages over the method used in earlier
structure – through design and communication – to career         studies where one’s position on the core-periphery
paths, such that prior to advancing to core administrative       continuum was determined based on a simple
duties, contributors should first serve on functional roles      quantification of his activity [48]12. Nonetheless, we
such as Border Patrol or Quality Assurance.                      propose that future studies employ a multi-method
Notwithstanding the recommendation above for                     approach, determining one’s role based on access
“professional” career paths, it may be useful to create – in     privileges, activity count, and additional recognition one
parallel - direct paths to leadership. Results from our study    receives (e.g. in Wikipedia: ‘barnstars’13 , service
suggest that this path may be attractive to many                 awards14 , and other personal awards15 [30, 33]). Second,
Wikipedians (as indicated by the large number of editors         our analysis of Wikipedia’s organizational structure has
who bypass Level to roles on their progression to becoming       focused on human agents, whereas automated software
an Administrator). A design implication would be to              bots, too, carry out important tasks [42]. We, thus,
develop tools that track contributors’ activities and offer to   propose that future studies provide a more complete view
them career guidance, including suggestions for roles and        of work organization by also analyzing the various tasks
functions that best match their profile.                         performed by bots. Third, we acknowledge that the role
                                                                 transition patterns reported in this study have been
Our findings are also relevant to managers in firms who
                                                                 influenced by the process by which the Wikipedia
explore new collaborative production strategies. For
                                                                 community introduced access privileges. For example, in
example, some technology companies participate in open
                                                                 the early stages of Wikipedia development, administrators
source software development [51], whereas others have
                                                                 were responsible for many of the functional roles;
adopted the practices of the open source movement for
                                                                 however, after the introduction of Level 2 privileges
their internal software development projects [50]. In a
                                                                 (2008-2011) many of these functions were delegated to
similar vein, many companies use wiki technology as a
                                                                 the Technical Administration, Border Patrol, Quality
knowledge management tool [2], and in particular for
                                                                 Assurance and QA Technicians roles. This implies that
developing a Wikipedia-like organizational encyclopedia
                                                                 the nature of Administrator role (as well as patterns of
[28], adopting (at least in part) the organic processes that
                                                                 transitions into this role) have changed over time, calling
typify wiki-based collaboration over the Internet. Insights
                                                                 for a future research that would re-examine role transition
from this study would be valuable to such firms, help
                                                                 patterns once roles stabilize. Fourth, in order to increase
structure the community of practice, organize work more
                                                                 the generalizability of our findings, in future research we
effectively, design career paths, and ensure the
                                                                 plan to expand the analysis and perform a larger-scale
community’s sustainability.
                                                                 evaluation (more participants, longer time period). In
CONCLUSION                                                       particular, the sample we employed excluded non-
Taking the results of this study as a whole, we make a case      member contributors and late joiners (becoming active
for investigating the functional organization of online          after 2007), and we call for future research that would
production communities. It is interesting to note that only      repeat our study on a more comprehensive sample. Lastly,
now researchers of online communities are turning their          in future research we plan to extend our investigation to
attention to the functional organization of production           other online communities, in particular those that differ
communities, given that this topic is at the core of the         from Wikipedia in terms of the motivational drivers, the
organizational literature [40]. Existing conceptualizations      governance structure, or the enabling IT platform.
of emergent leadership within peer production have               Moving the investigation beyond Wikipedia is of special
primarily focused on the dimension of authority (i.e.            importance, given that the distinct role-transition patterns
proximity to the community core) and prior studies have          we have identified – and in particular the importance of
largely relied on contributors’ activity for estimating the      the mid-level functional layer – may be relevant to only
position on the periphery-core continuum. We, however,           production communities that have grown beyond a certain
argue than an understanding of the formal functions in the       size. We call for future research that would explore the
community – as well as the way in which contributors             phase in a community’s life that call for the establishment
traverse these functional roles – is paramount for               of this kind of functional layer.
explaining how online production communities organize
their work.
                                                                 12
Given that this is only a first attempt to empirically             Preece and Shneiderman note that “…these metrics
analyze and quantify the process of role transition along        [employed in previous studies] only capture activity. The
participants’ career paths in online production                  development of more potent measures of efficacy in achieving
communities, our study has several limitations which we          personal and community goals would be a big breakthrough”.
hope to address in future research. First, in terms of           13 http://en.wikipedia.org/wiki/Wikipedia:Barnstars
                                                                 14
method, building on [3] we defined organizational roles             http://en.wikipedia.org/wiki/Wikipedia:Service_awards
                                                                 15 http://en.wikipedia.org/wiki/Wikipedia:Awards
You can also read